Adaptive Audio-Based Context Recognition

Abstract

Abstract-Context recognition is an essential aspect of intelligent systems and environments. In most cases, the recognition of a context of interest cannot be achieved in a single step. Between measuring a physical phenomenon and the estimation or recognition of what this phenomenon represents, there are several intermediate stages which require a significant computation. Understanding the resource requirements of these steps is vital to determine the feasibility of context recognition on a given device. In this paper, we propose an adaptive context-recognition architecture that accommodates uncertain knowledge to deal with sensed data. The architecture consists of an adaptation component that monitors the capability and workload of a device and dynamically adapts recognition accuracy and processing time. The architecture is implemented for an audio-based context recognition. A detail account of the tradeoff between recognition time and recognition accuracy is provided.

Analysis of Error-Agnostic Time- and Frequency-Domain Features Extracted From Measurements of 3-D Accelerometer Sensors

“…Comparatively, the research community has so far focused on three aspects of signal processing, namely, aggregation, sampling, and compression. Dargie et al [6] and Chao et al [7] employ a local maxima technique to reduce the network's traffic. Ganesan et al [8] propose data aggregation and compression mechanisms based on spatial interpolation of data and temporal signal segmentation.…”

“…Therefore, it is necessary to divide the time-series measurements into different, short-duration windows. In order to avoid frequency leakages at the two edges of each window, it is customary to overlap neighbor frames, as a rule the overlap is 25% to 50% [6], [19]. Afterwards, FFT is performed on each of these windows.…”

“…The resource consumption of these steps is discussed in more detail in [6] and [19]. Additional to these processes, we carry out frequency normalization using a hamming window, so that the frequency resolution for each measurement is optimized.…”

Evaluation of the reliability of RSSI for indoor localization

“…Some of these applications use location information to infer the activity of mobile objects, animals, or human beings [7]. For example, biomedical sensor nodes can be attached to the bodies of patients [6] and nurses [5] to monitor their activities; workers in disaster recovery scenes [18] and oil extraction and refinery areas [4] can carry sensing devices to avoid dangerous situations; mobile sensor nodes can also be employed to report or debrief soldiers the events encountered during a mission [16].…”

Why is context-aware computing less successful?

“…Moreover, additional context sources, both direct and indirect sources, were employed to enrich the knowledge of recognition tools [8]. Remarkable accuracies in context-recognition were reported to have been achieved.…”